A0547
Title: A unified framework for parameters estimation in finite mixture models
Authors: Sayan Mukherjee - Duke University (United States)
Yun Wei - Duke University (United States) [presenting]
Long Nguyen - University of Michigan (United States)
Abstract: For parameters estimation in finite mixture models, the minimum distance estimators and the denoised method of moments estimator are known to be minimax optimal. We provide a unified framework, including both estimators as special cases, and the unified framework could be applied to produce new estimators by choosing different classes of test functions. Theories are obtained under the unified framework and they extend the existing theories when specializing our framework to their cases.